- 14 Jul, 2005 24 commits
-
-
Chen, Kenneth W authored
I was going over the scsi I/O submit path, when sd_init_command construct the scsi command, this_count is already checked in the previous else if clause. Why does it need to check it again in the last else block? Patch to delete the spurious check. Signed-off-by:
Ken Chen <kenneth.w.chen@intel.com> Signed-off-by:
James Bottomley <James.Bottomley@SteelEye.com>
-
Chen, Kenneth W authored
scsi_init_io calls scsi_alloc_sgtable and then calls blk_rq_map_sg to initialize the scatterlist structure. blk_rq_map_sg() already memset the structure for every new segment. That makes the memset in scsi_alloc_sgtable unnecessary. Patch to delete the extra memset in scsi_alloc_sgtable. Tested on a x86_64 machine. Looks stable to me. Signed-off-by:
-
James.Smart@Emulex.Com authored
One of the issues we had was reverting the midlayers lun value into the 8byte lun value that we wanted to send to the device. Historically, there's been some combination of byte swapping, setting high/low, etc. There's also been no common thread between how our driver did it and others. I also got very confused as to why byteswap routines were being used. Anyway, this patch is a LLDD-callable function that reverts the midlayer's lun value, stored in an int, to the 8-byte quantity (note: this is not the real 8byte quantity, just the same amount that scsilun_to_int() was able to convert and store originally). This also solves the dilemma of the thread: http://marc.theaimsgroup.com/?l=linux-kernel&m=112116767118981&w=2 A patch for the lpfc driver to use this function will be along in a few days (batched with other patches). Signed-off-by:
-
Andrew Vasquez authored
Cleanup FC remote port registration. Due to the inherent behaviour (an immediate scan) of adding a 'target'-role-capable rport via fc_remote_port_add(), split the registration into two steps -- addition as unknown-type role, then use fc_remote_port_rolchg() with appropriate role (based on PLOGI/PRLI bits). This allows for a more cleaner rport->dd_data management as can be seen with the simplified qla2xxx_slave_alloc() function. Signed-off-by:
Andrew Vasquez <andrew.vasquez@qlogic.com> Signed-off-by:
-
Andrew Vasquez authored
Consolidate ISP24xx chip reset logic. Consolidate near-duplicate RISC reset logic from qla24xx_reset_chip() and qla24xx_chip_diag(). Also, after initiating a soft-reset, insure the firmware has completed all NVRAM accesses before continuing. Signed-off-by:
-
Andrew Vasquez authored
Add firmware version number to qla24xx_fw_version_str(). Original code was accidently trimmed during port. Signed-off-by:
-
Mark Haverkamp authored
The wrong sgmap structure is being assigned in aac_send_raw_srb. Signed-off-by:
Mark Haverkamp <markh@osdl.org> Signed-off-by:
-
Andrew Vasquez authored
Update version number to 8.01.00b5-k. Signed-off-by:
-
Andrew Vasquez authored
Correct maximum supported lun and target-id definitions. The driver uses command-IOCBs which support a maximum lun value of 0xffff -- correct #define to reflect the change. Also, remove superfluous MAX_TARGET definition. Signed-off-by:
-
Andrew Vasquez authored
Update copyright banner. Signed-off-by:
-
Andrew Vasquez authored
Firmware updates. Resync with latest 21xx firmware -- 1.19.25. Resync with latest 22xx firmware -- 2.02.08. Resync with latest 23xx/63xx firmware -- 3.03.15. Signed-off-by:
-
Andrew Vasquez authored
Code scrubbing. - Remove trailing whitespace from driver files. - Remove unused #defines and inlines. - Standardize on C comments (// -> /* */) Signed-off-by:
-
Andrew Vasquez authored
NVRAM id-list updates. Resync with latest NVRAM subsystem ID list. Signed-off-by:
-
Andrew Vasquez authored
Add OS initialization codes for ISP24xx recognition. Signed-off-by:
-
Andrew Vasquez authored
Add ISP24xx initialization routines. Signed-off-by:
-
Andrew Vasquez authored
Add ISP24xx ISR routines. Add appropriate glue-code for ISP24xx support -- this included generalizing some of the core handling routines (qla2x00_async_event() [pull-up retrieval of mailbox values] and qla2x00_status_entry()]. Fixup 2100/2300 ISRs to handle the new conventions. Signed-off-by:
-
Andrew Vasquez authored
Add ISP24xx IOCB manipulation routines. Add appropriate glue-code for ISP24xx support while manipulting IOCB packets. Add an ISP24xx specific 'start_scsi' routine due to command-type-7 layout changes. Signed-off-by:
-
Andrew Vasquez authored
Add ISP24xx flash-manipulation routines. Add read/write flash manipulation routines for the ISP24xx. Update sysfs NVRAM objects to use generalized accessor functions. Signed-off-by:
-
Andrew Vasquez authored
Add MBX command routines for ISP24xx support. Generalize several routines [qla2x00_load_ram_ext(), qla2x00_execute_fw(), qla2x00_verify_checksum()] to handle larger addressing space. Signed-off-by:
-
Andrew Vasquez authored
Generalize SNS generic-services routines. Consolidate completion-status checking while adding support for the ISP24xx. Signed-off-by:
-
Andrew Vasquez authored
Add ISP24xx diagnostic routines. Add function and structure definitions for the ISP24xx diagnostic firmware dump routines. Signed-off-by:
-
Andrew Vasquez authored
Add ISP24xx definitions. Add requisite structure definitions and #define's for ISP24xx support. Also drop volatile modifiers from device_reg_* register layouts as the members are never really accessed, only their offsets within the layout are used during reads and writes. Signed-off-by:
-
Andrew Vasquez authored
Add pci ids for new ISP types. Move old definitions in local qla_def.h file to pci_ids.h as well. Signed-off-by:
-
Andrew Vasquez authored
Factor-out ISP specific functions to method-based call tables. In anticipation of ISP24xx/ISP25xx support, factor-out ISP specific functions into a method-based call table. Signed-off-by:
-
- 11 Jul, 2005 6 commits
-
-
James Bottomley authored
At the moment, the transport class blindly tries to set things like QAS and IU, even if the drive won't support them. It's best not to annoy the devices like this and instead only set what the drive says is actually supported. Signed-off-by: -
Christoph Hellwig authored
now that we do normal PCI probing there's no need to keep a list of all HBAs. Rejections fixed up and Signed-off-by: -
Christoph Hellwig authored
always probe in bus order, avoid any reordering Signed-off-by: -
James Bottomley authored
Here's a tiny update that means we print the correct ASCII type information Signed-off-by: -
Andrew Morton authored
drivers/scsi/dpt_i2o.c: In function `adpt_queue': drivers/scsi/dpt_i2o.c:385: warning: unused variable `timeout' Signed-off-by:
Andrew Morton <akpm@osdl.org> Signed-off-by:
-
Andrew Morton authored
Use the macros consistently in ahd_linux_dev_reset(). If ahd_linux_dev_reset() really can be called with local interrupts disabled then yuk. Signed-off-by:
-
- 03 Jul, 2005 6 commits
-
-
James Bottomley authored
Apparently these are the only drives that try to negotiate IU and QAS at u160 speeds. The aic7xxx driver can't cope with this. The fix is to eliminate the IU and QAS setting routines. I've #if 0'd them out, just in case we ever get the sequencer documentation out of Adaptec, since we'd then be able to fix the driver. Signed-off-by: -
Linda Xie authored
Fix the problem in IBM VSCSI Client where the client doesn't send the information which is expected by the server. Signed-off-by:
Linda Xie <lxie@us.ibm.com> Signed-off-by:
-
Christoph Hellwig authored
This code pokes deep into EH internals and duplicates scsi_error.c code wrongly. It doesn't compile anymore in scsi-misc, so let's #if 0 out the code - the driver hasn't worked for more than five years anyway. Signed-off-by: -
James.Smart@Emulex.Com authored
Signed-off-by:
James Smart <James.Smart@emulex.com> Signed-off-by:
-
James.Smart@Emulex.Com authored
Update copyright notice text and include year 2005. Add Copyright notice for Christoph Hellwig to several files: lpfc.h lpfc_attr.c lpfc_els.c lpfc_hbadisc.c lpfc_init.c lpfc_mbox.c lpfc_mem.c lpfc_nportdisc.c lpfc_scsi.c lpfc_sli.c Signed-off-by:
-
James.Smart@Emulex.Com authored
Signed-off-by:
-
- 02 Jul, 2005 4 commits
-
-
James.Smart@Emulex.Com authored
Symptom - An unmapped node (initiator) that goes away in a situation such as cable pull, comes back as a mapped node. Fix - On ADISC completion, put a list on the mapped list only if it is a FCP_TARGET. Signed-off-by:
-
James.Smart@Emulex.Com authored
Add completion handler to the abort iocbs to close a hole where we could reuse an iotag. Signed-off-by:
-
James.Smart@Emulex.Com authored
lpfc_els_unsol_event() checks rjt_err to determine is LS_RJT should be sent. However, rjt_err was set to LSEXP_NOTHING_ELSE (which is 0) in cases where an LS_RJT should be sent, so rjt_err was never true. Change lpfc_els_unsol_event() to set rjt_err to 1 when LS_RJT should be sent. Signed-off-by:
-
James.Smart@Emulex.Com authored
Fix driver not seeing LP6000. Fix: add PCI id to the pci_device_id table and a short description for the HBA in get_hba_model_desc(). Also add a default clause to the switch statement that parses the various PCI ID's. Signed-off-by:
-
